About this project

Let me see is an interactive experiment that explores how a machine’s way of communicating shapes our perception of its intelligence. It uses a simple character-recognition system that describes what it sees, hesitates and asks users whether its answer is right.

Why OCR?

OCR (Optical Character Recognition) turns text in images into digital text. I chose it because we associate reading with intelligence. When we read, we do more than recognize letters: we understand and interpret their meaning. This led me to ask: can a computer’s act of recognizing text be considered intelligent?

How does this reader work?

This prototype reconstructs one part of OCR: visual comparison. It converts an image to black and white, isolates a character and resizes it onto a pixel grid. It then compares the shape with reference letters and digits, measuring how much they overlap.

The similarity scores and rules I set determine whether it gives an answer or remains uncertain. Typeface, contrast, rotation and image quality can affect the result. The percentages measure shape similarity, not the probability of a correct answer. The system compares patterns; it does not understand what a character means.

Why make it a conversation?

The system has already calculated its answer before it starts speaking. But its descriptions, hesitation and humor make it seem as though it is looking, thinking and gradually deciding. It asks users to confirm its answer and can offer another candidate, turning a calculation into an exchange.

I designed the site so that users experience this conversation first, then explore the comparisons behind the answer. This lets them encounter both the impression the machine creates and the mechanism that produces its result.

Reading and understanding are not the same.

Through this project, I realized that even simple things like hesitation and humor can make a machine seem more human. They can also make its process feel more complex than it actually is. The prototype does not gain understanding through conversation. Yet its language and responses can make a visual comparison feel like an act of thinking.

How a machine speaks changes how we see it. The machine’s underlying ability stayed the same. What changed was how it presented that ability and how we might perceive it. I want users to question whether the machine seems intelligent because it recognizes a character, or because it speaks like a person.